Color and Grayscale Medical Image Authentication and Self-Recovery Based on Integer-to-Integer Wavelet Transform

With the recent advancement in internet and communication technologies, the exchange of digital images has significantly increased, leading to the emergence of new services such as telemedicine. However, due to the availability of powerful image manipulation software, it becomes important to achieve authentication and anti-counterfeiting for medical images. In this paper, we propose a fully blind fragile watermarking model for the authentication and self-recovery of both color and grayscale medical images. Initially, the Integer-to-Integer Wavelet Transform (IIWT) is applied to the host image, and the midfrequency (LH, HL) and high-frequency (HH) subbands are divided into 3 \times 3 blocks. Subsequently, the authentication watermark is embedded into the HH subband, and two recovery watermarks are inserted within the LH and HL subbands using LSB substitution. During the extraction procedure, if the watermarked image is identified as altered, the recovery process takes place to restore the original image. Experimental results indicate that the proposed method produces high-quality watermarked images with an average Peak Signal-to-Noise Ratio (PSNR) of 51.92 dB and achieves perfect watermark extraction accuracy while maintaining high sensitivity against various attacks. Furthermore, the proposed scheme demonstrates highly accurate tamper localization with a precision of around 99% and achieves high recovered image quality when facing tampering. The source code of the proposed method can be accessed on GitHub at https://github.com/Riadh-Bouarroudj/Fragile-image-watermarking-with-recovery.